Engagement and Retention with Programmatic in the Post-IDFA Era – Virtual Session Recap

RevX discusses engagement and retention with programmatic for the post-IDFA era in an engaging session of the latest App Promotion Summit London. The session caters to all advertisers wanting to grow their apps with programmatic advertising and learn more about the current mobile ecosystem changes.

In this session, Martje Abeldt (Chief RevX Officer) gives you a historical perspective on the power of advertising. Since the main presentation topic is programmatic engagement and retention, this session focus on providing a new view for advertisers to help them make informed decisions on how to drive more efficient programmatic advertising. 

We address the impact of iOS 14.5 on the existing advertising landscape ahead of its full rollout during the session. All we learned about measurement options in a "device ID world" will not work anymore after iOS14.5. So, we look at what all advertisers need to know now and provide practical info on how advertisers can allocate budgets and structure their campaigns. Finally, we look into audience engagement targeting strategies that you can deploy successfully in the future for the non-device ID world.


On iOS14.5 impact to programmatic advertising

"What we know is the following. What can happen to mobile ad spend 2020 looks roughly like this: 250 billion. If we split this (spend) between iOS and Android, you see it's roughly 50/50. So what's in question is the proportion of what has been spent last year. Will this be spent the same way, or will it be spent differently? Will advertisers spend more? Ultimately, for iOS devices, I have users that I want to address: users I wish to acquire, reengage and target with branding campaigns to stay top of mind. So, all of this persists. 

What stops persisting is the mechanics used to address these users and the money that was being spent based on these existing mechanisms. 

What happens with Android

"While on Android, where the majority of users is available on a deterministic basis through a device ID, we can expect that the (advertising budget) amount will rise. The techniques and the ways how the money is being spent on Android will probably continue to be the same."

"We're going to be somewhere between campaigns where we can apply deterministic tracking, and others, where we can apply something like probabilistic tracking, is going to be a mix."


"The time(frame) when you can get information can be very, very long, depending on how you define the (conversion value) list. Ultimately, this leads to much confusion, right? So you have to understand how the SKAdNetwork works. 

Everybody has to register there. It's monopolizing the space; you have to rely only on one source of truth. Then, you have to configure all your events. This is highly difficult. 

I think nobody is fully ready. That means if (iOS14.5 changes) rolls out immediately, scale is a problem, at least until this confusion settles, and until Apple improves the system, which will probably take another six to 12 months."

On Fraud

"We have another problem here, which is fraud. I listed down five very basic mechanisms on how to prevent fraud. There are many, many more. But the complexity of SKAdNetwork opens the door widely to fraud: fraud from the supply side and fraud from the vendor side. 

Ultimately, the advertiser is left with fewer data points they can leverage. And of course, for example, these two are dependent on device ID, so only with the device ID and with timely measurement of events I can optimize for post-install engagement and understand whether there is no action after the install. 

Because it's a fraudulent user, databases I have created to list device IDs, which are fraudulent, I cannot use either. So I'm losing out on fraud prevention. That means the likelihood that advertisers are losing budgets because of fraud increases with a new way Apple is implementing and imposing SKAdNetwork. 

So what are we left with ultimately? What do we have to do? Well, traditional uplift tests don't work on LAT because they're all dependent on the device ID. There was a control group and a test group, regardless of whether it's intent-to-treat, placebo ads, ghost bids, ghost ads... everything was dependent on device IDs to determine who falls into which group and how I would measure the performance of each of these groups. So what you've learned in the past doesn't work anymore.

What works is something that is not new at all, but it hasn't been applied at scale to advertising yet. At least, I haven't heard from any advertiser that they're doing it, which is something like econometric time series analysis."

On measurement

"If you think about these (measurement) models, how do you actually build a system? How do you create a system for yourself? Advertisers will have to do this for themselves because every advertiser likes different products and works with different channels. So you have to find a mechanism where you can create treatment and control groups that you can compare against each other. And then you have to find a modus operandi where you say: "Okay, what are my baselines? Which channels do I test? And how do I test them? Because you cannot only test one channel, and then another one, but maybe you're testing several channels at once."

"I think everybody has been settling on the last-click attribution work because it's comfortable, it's fast, it (allows you to) take fast action, you have something to show. But (this method) hasn't been measuring true incrementality. 

We've been settling with a compromise. And I think it's understandable because there's also a challenge of timing, a challenge of complexity. After all, anybody who wants to do true incrementality testing at scale for all the channels needs to set this up at once. So, there will be significant labor involved to do this. Once it's done and the system is up, I think it creates a lot of benefits because you will truly understand how your budgets are performing."

How to look at audience engagement targeting strategies and creatives

"This is the way how we should think about creative innovation: we should make ads more versatile so they can address all of these people because every impression that goes out is potentially addressing any of those use cases."

From the late Bill Bernbach: "Nobody comes the number of ads you run; they just remember the impression you make." It is incredibly important. 

"We have to make good impressions in the future as an advertiser; this is one of the core initiatives I think we should undertake: to make advertising much, much active again. And this goes along with: quantity doesn't count as much as quality. 

So, does it count that I'm shooting all 10 of these impressions out to users? And they're all the same? And it's kind of boring. It's kind of repetitive. Or should I focus on designing fewer creatives, maybe larger, more engaging interactive formats, to make sure I make an impression on the user? Because, as we know, branding works, right? And branding also leads to UI or engagement. So again, it's all in one. And we have to make good impressions in the future."

On supply signals

"The publishers need to make more supply signals available to us than they're doing now. This is the work they have to do. I've asked around a little bit, but it doesn't seem that anybody has moved in big steps. So I guess the rollout of iOS 14.5 needs to happen first. And then, we'll slowly, gradually move into a stage where probably the information in the first place will be more homogeneous. That means instead of five to 10 to 15 data points that we're using today that are maybe homogenous; maybe we'll be getting like 20, 25, 30 data points that are existing in a homogeneous matter so I can use them for targeting. 

In the future, maybe we're adding more qualitative information about the impression. Everybody speaks about contextual data, but when we go deep, and we try to understand how much context real data is there actually, it's still very superficial."

In sum, how can advertisers proceed to be prepared for the iOS14.5 changes?

"To wrap this up, I would advise everybody, take a step back, right? I think everybody has been in this for months and months in tonnes of webinars and tonnes of presentations, (to figure out) how to implement SKAdnetwork and so on, do your homework. But, let's take a step back. Let's lose the panic and think about advertising, and what is at the core of advertising? What makes sense? What makes advertising really, really good? What are the goals of your brand? We all have a brand; every app is a brand. Now, how do we bring this to the forefront? How do we make it attractive, memorable right?

Then, understand the impact of iOS 14 and LAT on your business versus the whole world. Maybe you find out that you can tweak it, iterate on it for your own business to minimize the impact or find new solutions for your business as well. 

Then, evaluate measurement options. So keep in mind, the measurement options that you knew from the past will not work in the future, except if you blindly follow SKAdnetwork and everything that comes out of there. But I think it seems to be clear that you probably won't be able to rely on it. Because, as you see, it's too complex, it's too "un-ready", and it bears many risks, respect to fraud.

Then, test media mix optimizations with your channel partner. You need your channel partners to collaborate and set up a system that enables you to do this A/B testing the setting up these control groups, checking backend data, and doing this over and over again, in a simplified manner. 

So if you don't find a setup where you can do this fast and simple, you won't do it yourself. You want to think about how you are testing your media mix properly, so you are ready for long-term success. 

Get creative when thinking about addressing your users. These users are consumers. And we should all be thinking consumers first. Advertising was maybe not thinking consumers first. They were thinking advertisers first, getting out there first, hammering down on users with as many impressions as possible. And that's not what we should do.

This is a good moment to change how we operate and become more creative again, ultimately making advertising much more enjoyable for our consumers. That means making a lasting impression for your brand that helps you to drive long-term value for both the consumer as well as your company."

Marta Gaspar

Marta is a digital marketer and a content strategist with 11 years of experience turning ideas into prose and campaigns. She likes to stay updated on the mobile marketing scene and often writes about AdTech.